Efficient Event-Driven Simulation of Large Networks of Spiking Neurons and Dynamical Synapses

Abstract

A simulation procedure is described for making feasible large-scale simulations of recurrent neural networks of spiking neurons and plastic synapses. The procedure is applicable if the dynamic variables of both neurons and synapses evolve deterministically between any two successive spikes. Spikes introduce jumps in these variables, and since spike trains are typically noisy, spikes introduce stochasticity into both dynamics. Since all events in the simulation are guided by the arrival of spikes, at neurons or synapses, we name this procedure event-driven. The procedure is described in detail, and its logic and performance are compared with conventional (synchronous) simulations. The main impact of the new approach is a drastic reduction of the computational load incurred upon introduction of dynamic synaptic efficacies, which vary organically as a function of the activities of the pre- and postsynaptic neurons. In fact, the computational load per neuron in the presence of the synaptic dynamics grows linearly with the number of neurons and is only about 6% more than the load with fixed synapses. Even the latter is handled quite efficiently by the algorithm. We illustrate the operation of the algorithm in a specific case with integrate-and-fire neurons and specific spike-driven synaptic dynamics. Both dynamical elements have been found to be naturally implementable in VLSI. This network is simulated to show the effects on the synaptic structure of the presentation of stimuli, as well as the stability of the generated matrix to the neural activity it induces.

Simulation of networks of spiking neurons: A review of tools and strategies

“…One iteration consists in In the simple case of identical transmission delays, the data structure for the queue can be just a FIFO queue (First In, First Out), which has fast implementations (Cormen et al 2001). When the delays take values in a small discrete set, the easiest way is to use one FIFO queue for each delay value, as described in Mattia and Del Giudice (2000). It is also more efficient to use a separate FIFO queue for handling random external events (see paragraph about noise below).…”

Managing Burstiness and Scalability in Event-Driven Models on the SpiNNaker Neuromimetic System

“…To improve performance, recent software tools have turned to event-driven computing [7,34,58]. However, conventional sequential computers do not usually have direct hardware support for event-driven applications, and thus most event-driven simulators actually run an emulation by using a small timestep, recording events in an event queue, and updating all processes dependent upon the events in the queue at the appropriate timestep [32,53].…”

Independent Variable Time-Step Integration of Individual Neurons for Network Simulations

“…(Makino, 2003;Mattia & Del Giudice, 2000;Watts, 1994) However, these networks have been restricted to use with artificial cells which permit analytic solution or approximation of cell states based on values at an arbitrary prior time. In such a network, cell states are calculated at the time of event receipt based on values determined at the prior event.…”

“…To handle an event, the simulator updates the states of follower cells and places on the queue any new events generated by these followers. Since many practical simulations involve only a small set of event delay times, the need for O(logN c ) queue sorting is avoided and the event-queue can make use of efficient algorithms such as the O(1) algorithm presented by Mattia and Del Giudice (2000). Similar to the above, a network in Neuron can be constructed entirely of event-driven artificial cells.…”